基于空间尺度寻优的公共信息标志的主动搜索

Active Search and Detection of Public Information Sign Based on Prior Space Scale Information

为了使"福娃机器人"能够在机场候机楼中自主导航,以公共信息标志为导航路标,研究公共信息标志的搜索与检测问题,提出了一种基于视觉时空多尺度信息的标志搜索与检测方法。利用公共信息标志形状的先验信息,在空间大尺度下对公共信息标志快速搜索。当公共信息标志在图像上呈现模糊、标志部分信息缺失、标志粘连等现象时,根据时空相关性自动进行空间尺度变化、寻找最优空间尺度完成公共信息标志的二次搜索与检测。实验结果证明了该方法的快速性和有效性。

To solve the autonomous navigation problem of ＂Fuwa robot＂ in airport terminal, taking public information signs as navigation signs, the methods of searching and detecting public irformation signs are studied and a new method based on multi- scales and spatiotemporal information is put forward. Using the prior information of public information signs, public information signs are searched fast in a big space scale. When the phenomena of blur, lack and conglutination in image occur, imaging position and space scales arc changed automatically on the basis of spatio- temporal relativity and the best space scale is found to .search and detect the public information signs successfully. Experiments show that the method is rapid and efficient.